Stereo-PropCam (final silicon: 1cog, 8 pins, 550FPS, 360MHz ...480 clocks left:)

No pictures just yet. I need to get the data onto an SDcard... and there are some things standing between here and there.
I think with 400MHz, we might squeeze another PropCam into that cog.

Comments

  • 550 frames per second!? Is that meant to be 50 FPS? Sounds neat anyway.

    You do realise that 360 MHz will slowdown, or maybe crash, if it gets warm. You'll need a chiller for 400 MHz. I suppose an appropriately operating Peltier would do it.

    We have the vastness of the internet and yet billions of people decided to spend most of their time within a horribly designed, fake-news emporium of a website that sucks every possible piece of personal information out of you so it can sell it to others. And they see nothing wrong with that.
  • 550 frames. Global shutter. Perfectly stable. Synchronized right down to the pixel.
    Works with just one PropCam too!

    A GREAT CHRISTMAS GIFT FOR YOU AND THE WHOLE FAMILY

    It works at 250MHz... not quite the frame rate but close. You just have to tweak one parameter.

    The code does work unmodified with one PropCam, but then you can't use one pin at PropCam_base +6. Next time I'll show the lines to change if you just have to use that pin:)


    It has run for over 24 hours without a crash at 360. Silicon lottery I guess.

    Peltier coolers are made here in town, but I was thinking of a shot glass with some ice.

    I'm thinking of reworking the logic a little. It doesn't make sense to do it for just two PropCams, but for an array, there is a whole different logic possible. I'm pretty excited about it. Have to get HyperFlash working right get two boards talking...mmmm.

    I have trouble sleeping at night... can't stop thinking about all of the possibilities.







  • jmgjmg Posts: 14,150
    rjo__ wrote: »
    It works at 250MHz... not quite the frame rate but close. You just have to tweak one parameter.

    What frame rate does 250MHz give ?

  • rjo__ wrote: »
    It has run for over 24 hours without a crash at 360. Silicon lottery I guess.
    It's all down to die temperature. If overall the prop is not working hard then it won't warm up much at all. And if ambient is low you'll have a little more wriggle room. What's more, as Chip has observed, the PLL is the primary limiting factor so crashes are unlikely. It'll just go a little slower than 360 MHz instead.

    We have the vastness of the internet and yet billions of people decided to spend most of their time within a horribly designed, fake-news emporium of a website that sucks every possible piece of personal information out of you so it can sell it to others. And they see nothing wrong with that.
  • evanhevanh Posts: 8,391
    edited 2019-11-30 - 06:16:29
    From memory, 400 MHz is a lot more touchy. It does tend to crash if the temperature is low enough to being close to achieving 400 MHz. Temperature has to be kept well below that to get reliable execution and solid 400 MHz.
    We have the vastness of the internet and yet billions of people decided to spend most of their time within a horribly designed, fake-news emporium of a website that sucks every possible piece of personal information out of you so it can sell it to others. And they see nothing wrong with that.
  • jmg

    I didn't measure it, but with the P123 at 80MHz I was getting close to 350FPS for a single prop... I think. The pixel readout is serial at 10MHz... then there is a delay in handling the data until a new frame is called for. So... at 250MHz... fairly maxed out. I just can't say for sure.

    I've been think about this multiple Propeller thing... I can do a lot with what I have, but I think there is going to be a lot of bit banging involve. When it comes to using those SmartPins as God intended, I think we are going to need a common clock... and you seem to be the clock king around here:)

    Two of the developments that I think are highly significant is Brian's work demonstrating that he can drive HDMI at 24 bits... for an image residing in hubram. The second is Rayman, who demonstrated that if you get the data fast enough the buffer in hubram can be about the size of a line... For a P2x4 that keeps the display P2 with a clean slate in hubram… and you know how these guys love their hubram.

    Just saying;)


  • rjo__rjo__ Posts: 2,076
    edited 2019-11-30 - 06:50:47
    evahn

    I don't know for sure. My understanding was that there was a PLL limit, that we don't really know have fast the core logic can run. I crash around 375MHz no matter how little I ask the P2 to do. It is immediate... So, I don't know if that is heat or the PLL limit on my machine

    If the core logic will run at 400MHz or above, the application would dictate whether you would ever need to go that high.

    I don't think cost is going to be an issue for some application areas. Students and hobbyists are cost sensitive.
    But when you need to use a P2 because no other solution makes sense or is even possible, that's another ball game.
    The P2 has some pretty interesting specs. Exactly where that puts it in the world of commerce.. we don't know yet.
    But I wouldn't sneeze at it.

    I'm perfectly happy at 320MHz or even 250MHz. Total MIPS, running out of cogs and pins are my biggest issues.

    I have a plan for that.





  • I'm off to bed.
  • evanhevanh Posts: 8,391
    edited 2019-11-30 - 07:54:12
    rjo__ wrote: »
    My understanding was that there was a PLL limit, that we don't really know have fast the core logic can run. I crash around 375MHz no matter how little I ask the P2 to do. It is immediate... So, I don't know if that is heat or the PLL limit on my machine
    Sounds about right. I found they are very close and is definitely temperature related - which is work load dependant too. For some situations the PLL is the limit (it just runs slow without crashing), for other situations the logic is the limit (it crashes). There may also be a small differing bias on each chip as well but temperature seems to be the biggest factor.

    We have the vastness of the internet and yet billions of people decided to spend most of their time within a horribly designed, fake-news emporium of a website that sucks every possible piece of personal information out of you so it can sell it to others. And they see nothing wrong with that.
Sign In or Register to comment.